Deliberative forums in fragile contexts: Challenges from the field

Abstract

What does it actually take to run a deliberative mini-public (DMP)? This article unmasks the taken-for-granted assumptions of one of the most popular democratic innovations in the past decade by applying their design features in fragile political contexts. Drawing on 3 years of fieldwork in communities recovering from armed conflict and police brutality in the Philippines, we identify foundational challenges to DMPs in relation to their core design features: sortition and long-form deliberation. Our goal in documenting these challenges is not to devalue DMPs but to spotlight limitations that cannot be overcome despite the best intentions. In so doing, this article hopes to advance an honest conversation about what these DMPs can reasonably achieve, as well as the harms they can potentially cause, especially in societies where precarity is the norm rather than the exception.

Introduction

In recent years, sortition plus deliberation have been portrayed as the winning formula for citizen engagement. In 2020, the Organisation for Economic Co-operation and Development (OECD) published a report documenting the explosion of deliberative mini-publics (DMPs; or representative deliberative processes) in Europe, North America, and Australia (OECD, 2020). Part of the allure of these democratic innovations is their design features that bring the principles of deliberative democracy to life (see Curato et al., 2021). To realise the principle of inclusiveness, DMPs use sortition or random selection to bring together a microcosm of society that reflects the diversity of the wider population. To enforce the principles of equal voice, respect, and reasoned discussion, DMPs use a combination of small group and plenary deliberations, typically with the guidance of trained facilitators. To reduce asymmetries in information, participants have access to information packets and expert testimonies to inform their deliberations. These design features set DMPs apart from forms of citizen engagement that invite ‘the usual suspects’, or political practices characterised by confrontation, divisive rhetoric, spin, propaganda, and misinformation. Ultimately, DMPs aim to generate considered judgement based on inclusive, informed, and reasoned discussion.

But what does it actually take to run a DMP? What are the taken-for-granted assumptions of process designers when sortition is used as a recruitment strategy and long-form deliberation as a mechanism of citizen participation? Whose realities are these assumptions based?

This article presents lessons for designing and implementing DMPs and forums for citizen deliberation, more broadly, based on empirical reflections from the field. Drawing on our 3-year, multi-sited fieldwork in the Philippines, we document our experiences in attempting to design and implement DMPs in communities recovering from armed conflict and police brutality. Focusing on fragile political contexts allows us to lay bare the prevailing and often unchallenged assumptions about designing DMPs. We catalogue the structural constraints and contingencies we faced in recruitment and programme design, as well as the lessons we learned from letting the rhythms of everyday life shape the design of citizen deliberation. Our goal in documenting these challenges is not to devalue DMPs but to spotlight limitations that cannot be overcome despite the best intentions. In so doing, we hope to advance an honest conversation about what these forums can reasonably achieve, as well as the harms they can potentially cause, especially in societies where precarity is the norm rather than the exception.

Our article is structured in three parts. First, we revisit the normative assumptions underpinning the design of DMPs and the theoretical and empirical critiques against these forums. In the second section, we introduce our cases and the justification for their selection as bases for empirical and theoretical reflection. We situate our positionalities as researchers who have actively promoted deliberative forms of citizen engagement in the Global South while taking a critical stance against template-type thinking when designing these forums. The third section focuses on foundational challenges against DMPs that we encountered in our fieldwork: (1) people’s mistrust of demographic data on which sortition is based, and (2) the rejection of the temporal demands of long-form deliberation. In each section, we uncover the taken-for-granted assumptions we had when designing DMPs and the design adjustments we made upon realising that our assumptions did not hold. We conclude the article by identifying the lessons that can be learned from our reflections beyond cases of extreme hardship and pathways for future research.

Citizen engagement’s ‘flavour of the month’

‘Flavour of the month’ was how Graham Smith (2019) described citizens’ assemblies in an interview for the website Open Democracy. Reflecting on his decade-long work on democratic innovations (see Smith, 2009), he finds that the design features of citizens’ assemblies make it a ‘different kind of beast’. Bringing together 100 people representing the microcosm of society to engage in rigorous deliberations spanning from 4 to 6 weekends distinguishes citizens’ assemblies as forums embodying virtues that other forms of citizen engagement do not. Indeed, for advocates of citizens’ assemblies – and DMPs more broadly – its core design features make a difference.

Citizens’ assemblies, citizens’ juries, Citizens’ Initiative Reviews, and Deliberative Polling^TM are some examples of DMPs. Each of these examples has specific design features, although all are designed to fulfil the functions of a ‘democratic mirror’ and ‘epistemic filter’ (Lafont, 2019: 95). DMPs are designed to serve as a democratic mirror by generating a representative sample of the broader population through random selection or sortition. For some normative theorists, random selection upholds the principle of political equality by advancing ‘the fundamental belief in the basic equal capacity of all citizens to participate in discussions about complex policy issues’ (Leydet, 2019: 355; also see Landemore, 2013). Some DMPs use pure random selection to invite participants, but many use quota sampling via targeted recruitment to overrepresent voices not typically heard in political decision-making. Meanwhile, as an epistemic filter, DMPs are designed to create ‘incentive structures’ that encourage participants to deliberate by scrutinising expert evidence, weighing different perspectives, and engaging with different views with respect and reciprocity before generating collective recommendations (Setälä and Smith, 2018: 304).

DMPs are not without their critics. Long before the so-called ‘deliberative wave’ in OECD countries, the literature on political sociology, science and technology studies, and participatory design have expressed scepticism about the design logic underpinning DMPs (see Breher et al., 2017; Chilvers, 2008). For the purposes of this article, three critiques are worth identifying.

First, critics find DMPs to be a top-down ‘elite-led form of participation’ designed by ‘democratic engineers’, which crowds out organic forms of civic life (Blaug, 2002: 112; Bua and Bussu, 2021: 217). Critics argue that the core design features of DMPs are overly prescriptive, leaving little room for eventfulness. Facilitators, for example, keep participants focused on the task – whether the task is collating questions for experts or generating recommendations – while participants’ contributions judged off-topic or outside the remit of deliberations are respectfully acknowledged and then shelved. In this case, what counts as authentic deliberation is a ‘sanitised’ and ‘domesticated’ process of participation, one where ‘deliberation is managed expertly’ by efficiently moving participants to discover their common ground (Blaug, 2002: 112; Bua and Bussu, 2023: 1; see Polletta, 2020). For some political leaders, informal interactions with their constituents behind the scenes, in everyday settings like the local market or after an event, are much more preferable than participatory processes like DMPs for they see them as ‘staged, over-structured, formal, and unreal’ (Hendriks and Lees-Marshment, 2019: 607).

Second, DMPs are accused of promoting off-the-shelf solutions without serious attention to place-based considerations (Christensen and Grant, 2020). The particular experience of DMPs in OECD contexts is often universalised and applied in other societies that have specific political sensitivities and cultural sensitivities. As Jan-Peter Voß and Nina Amelung (2016: 762) put it, ‘design knowledge became trans-local, it was produced to be generic, mobile, and relevant across different contexts’. One lesson DMPs can learn from the explosion of participatory budgeting around the world is that the success of a democratic innovation is dependent on ‘the history and particularities of its place of origin’ (Von Busch and Palmås, 2023: 54). Participatory budgeting’s rise and fall in its birthplace, Porto Alegre, are closely tied to the ascendancy of Partido dos Trabalhadores (PT) or the Workers’ Party which forged an alliance between labour groups and social movements with municipal and state governments, while its decline was also associated to the waning power of PT (Melgar, 2014). This example serves as a reminder to DMPs – that its design cannot be exported wholesale without enabling conditions within the broader political context.

Finally, DMPs are critiqued for being exclusionary. DMPs are a demanding form of political engagement. Participation requires time, effort, epistemic resources, and psychological dispositions to endure hours of learning, deliberation, and reflection. As Kevin J. Elliot (2023: 4) argues, making political participation more difficult ‘locks busy people out of democracy’ and creates ‘subtle yet effective patterns of exclusion’. Here, Elliot is not referring to high-flying corporate executives who have packed calendars, but disadvantaged communities who are time poor due to caring obligations or working long hours to make ends meet. Feminist scholars have long argued that the unfair division of domestic labour disproportionately burdens women with caring responsibilities, reducing their political participation opportunities (Young, 2001).

Several of these criticisms have been addressed as DMPs evolved. Some have started loosening the programme of DMPs to give participants the power to chart the course of the deliberative process instead of concentrating control on process designers. It is increasingly common for DMPs to give participants the power to identify experts they want to hear from or to invite other stakeholders whose perspectives were missing in the room. Some citizens’ assemblies took longer than planned in response to the demands of Assembly Members for more time to flesh out their recommendations further. In Europe, there have been calls to mandate the right for citizens to take a leave from work to attend a citizens’ assembly to incentivise inclusion further. These, among others, signal the openness of process designers, as well as authorities commissioning these processes, to learning in practice to realise the democratic and epistemic potential of democratic deliberation.

Research methods

While DMPs can be improved through design tweaks and policy reforms, we must ask questions about their foundational assumptions. Assumptions, by their very nature, are taken for granted. They are presumed to be routine and mundane and, therefore, offer plausible premises on which our thoughts and actions can be based. Taken-for-granted assumptions, however, are not innocuous. When we ‘draw explicit attention to what we implicitly assume’, we uncover power structures that shape the conduct of everyday life (Zerubavel, 2018: 8). In the case of DMPs, the challenge is to lay bare these taken-for-granted assumptions, ask whose realities these assumptions are based on, and whose lived experiences are made to fit premises that do not correspond with how they live their lives. This line of enquiry facilitates a grounded understanding of what it actually takes to run a DMP. It exposes the hidden costs of design and participation that are often left invisible but nevertheless consequential for how process designers and participants experience democratic deliberation.

Our investigation begins with a multi-sited fieldwork in two communities in the Philippines from 2017 to 2019. Our goal was to conduct a pilot test on the ways in which DMPs can support communities in charting the course of their own recovery amid trauma and deep division. Drawing on the normative theory of deliberative and agonistic democracy, we considered it valuable to curate spaces ‘for sustained dialogue about and across difference’, such that communities can build relationships, learn about each other’s interests and considerations, and envision a shared future (see Dryzek, 2005; Maddison, 2015: 1023). We were inspired by the empirical work of Jürg Steiner et al. (2017: 20) on deliberative dialogues in post-conflict Colombia and Bosnia Herzegovina as well as favelas in Brazil experiencing ‘warlike situations’ due to drug-related violence between the police and residents. We wanted to apply the lessons we learned from these empirical studies in own country, the Philippines, that has its own history of armed conflict and drug wars, but also has a dense network of civil society groups and activist movements that generate vibrant and contestatory discourses in the public sphere (see Curato, 2022 [2021]; Quimpo, 2005).

Case selection

Our first case study is the Islamic City of Marawi on the island of Mindanao in Southern Philippines. ¹ In 2017, 5 months of air strikes and ground combat between the Philippine military and ISIS-inspired fighters forced 98% of Marawi’s residents to evacuate and temporarily settle in shelters outside the city. War correspondents compared the scale of the devastation to Aleppo in Syria and the nature of the conflict to Mosul in Iraq (see Fonbuena, 2020). We first went to the field a year after the fighting broke out to understand the lived experiences of affected communities. Most families remained in temporary shelters. Some protested the national government’s lack of public consultation on the city’s future. Marawi is an illustrative case where affected communities felt excluded from collective decision-making, making it a fitting empirical context to design a forum for inclusive deliberation.

Our second case study is a village in Quezon City on the island of Luzon in Northern Philippines. In 2016, then-President and populist strongman Rodrigo Duterte announced a violent crackdown against the drug trade and warned the nation that the drug war would be bloody (see Evangelista, 2023). Human rights groups estimated 12,000–30,000 drug-related killings by the police and vigilantes (Human Rights Watch, 2021), while the chief prosecutor of the International Criminal Court found there was ‘reasonable basis to believe that the crimes against humanity of murder, torture, and the infliction of physical injury and mental harm’ had taken place (Bensouda in Gutierrez, 2020). Our field site was a densely populated low-income community where residents make a living by reselling recycled scraps like wood and metal, and working as domestic or service staff in gated communities and commercial centres nearby. An investigative report by Reuters identified the police precinct in this community as ‘the most lethal’ in Duterte’s drug war, while the British newspaper The Telegraph described the community as a ‘killing field’. While we are generally sceptical of foreign media exoticising news stories from the Global South, we affirm the characterisation of our field site as one that has borne the brunt of the drug war. Like Marawi, Quezon City illustrates a case of a challenging communicative environment, but the kind where neighbours turned on each other because of a divisive and violent state policy.

Approach to designing citizen deliberation

Our research began by acknowledging the critiques against DMPs as top-down, decontextualised, and exclusionary forms of participation. We viewed the design features of DMPs, including sortition and facilitated deliberation, as starting points for a conversation, not an inflexible set of prescriptions of ‘how to do deliberation’. We were cognisant of our positionality as scholars and advocates of deliberative forms of citizen engagement, particularly our curiosity to apply the design features of DMPs in fragile political contexts. We also recognise our privilege as academics with job security living in cities with relative safety. This privilege had admittedly shaped some of our taken-for-granted assumptions when designing DMPs in both communities.

Recognising the partiality of our perspectives, we placed the practices of attentive listening and partaking at the centre of our work when designing DMPs (see Curato, 2019). We found these practices helpful in carrying out an ethical disposition to process design that was respectful of local contexts, mindful of complex power relations, and careful not to cause harm to already vulnerable communities. We started our research by doing several field visits over 2 years (2017–2019) to ground ourselves in the rhythms of affected communities’ everyday lives. We attentively listened to the anxieties of our research participants, many of whom do not have confidence in expressing their views in public due to fear and trauma. We conducted one-on-one interviews and informal group chats in the homes of some of the residents to develop our sensitivity in identifying which grievances people openly share, the worries they talk about in hushed tones, and topics they simply do not raise. This provided valuable insight into framing the topic of deliberation, as well as developing a process plan that is sensitive to the emotional and psychological toll of speaking and listening in a collective context. We have also partaken in affected communities’ everyday lives to get a sense of activities that they find valuable. Our frequent visits established our relationship with the community as committed, non-partisan brokers of conversation, not predatory researchers or agents of state surveillance (see Reinharz, 2017). This was especially important in communities that feel emotional vulnerability because of their lived experiences of grief and uncertainty.

Lessons from the field

After 2 years of field visits and conversations with community members in Marawi and Quezon City, we shifted our roles from embedded social science researchers to process designers of citizen deliberation. In our debriefing sessions, we put these roles in conversation to facilitate reflective practice and question our taken-for-granted assumptions that may be informed by epistemic biases based on our academic training and prejudices based on lived experience (Thompson and Pascal, 2012). We took our respective notes from these debriefing sessions, which served as the primary source of reflective data for this article.

One outcome of our reflective practice is the decision to make significant departures from DMP’s core design features, such that we ended up using the looser label ‘deliberative forum’ rather than a DMP to describe our process. This, to us, was not an indication of failure or lack of rigour. Instead, it signified our efforts to let our participants’ everyday experiences shape process design instead of imposing our preconceived notions of how inclusive deliberation should be structured. In the following sections, we discuss the taken-for-granted assumptions that were challenged in our fieldwork and the design adjustments we made to uphold the principles of democratic deliberation suitable to the contexts of the communities we were working with.

Mistrust and suspicion of demographic information

Participant recruitment is one of the most challenging phases of organising DMPs. For DMPs to serve their function as a ‘democratic mirror’, process designers often begin by sending an invitation to a randomly selected group of people drawn from databases such as the National Postal Service or the electoral roll and, in some cases, through cold calling or sending text messages (see Sortition Foundation, n.d.). From those who responded positively to the invitation, a random selection of people is drawn to generate a microcosm of the broader population based on age, gender, a proxy of socio-economic status such as occupation, postcode, or educational attainment, and, in some cases, attitudinal data. This process, arguably, is one of the most technical, if not opaque, aspects of putting together a DMP, although some organisations have made their open-source software for sortition publicly available for transparency and public scrutiny (see Flanigan et al., 2021).

There is increasing discussion in the academic literature and within the community of practice about the challenges of designing and implementing sortition. Questions about the integrity of databases have been raised, for incomplete population registers exclude specific groups of people, such as temporary migrants, refugees, and undocumented workers, among others (Curato et al., 2021; Gąsiorowska, 2023). Moreover, although sortition may give everyone a fair shot at being invited, the capacity or interest to participate is unevenly distributed. DMPs typically have a response rate of 5%, raising questions about the profile of people these forms of citizen engagement typically attract (see Redman et al., 2023).

We encountered similar observations in our field research, although our concerns about sortition are more foundational than issues on incomplete population registers and low response rates. Our fieldwork laid bare the need to problematise the production of demographic information on which participant recruitment is based (see Cobham, 2020). We observed how lists containing demographic information are not innocent tools to enable citizen participation via sortition but are political constructs that disadvantaged communities find suspicious. This observation warns of the democratic and social harms DMPs can cause when databases or population registers are uncritically used.

Our fieldwork in Marawi started with an unproblematic relationship with demographic information as the basis for sortition. We started designing our DMP by meeting with the community’s gatekeepers, in this case, the leader of displaced communities in a particular resettlement site. The community leader provided the list of internally displaced people living in the temporary shelters where we recruited participants for our DMP. They ² volunteered to take charge of recruiting participants at random, following our brief of selecting 10–15 people with diverse backgrounds, with 50% men and 50% women. They were successful in putting this group together, except that there were no young people who accepted the invitation.

Despite the good reputation of the community leader, some community members were suspicious about the list on which participant selection for the deliberative forum was based. In crisis-affected communities, population registers containing demographic information are contentious political constructs. These registers (or ‘lists’) can set the parameters of aid distribution, spelling the difference between people who qualify for aid and those who do not. For example, we heard mothers contesting the allocation of food packs according to ‘households’. ‘Household’ as a unit of aid distribution presupposes recipients to be a nuclear family of six, while in practice, a household includes extended kin. Lists become even more problematic when they are used to assert authority, as in the case of aid workers shutting down attempts of families to negotiate a larger share of food packs because ‘that’s not what the list says’. In this context, population registers are experienced as an instrument of exclusion, if not humiliation.

These lists are further politicised because they are used as instruments for patronage. In the Philippines, various studies have documented the selective distribution of aid to families that supported the political party in power, while families that supported the opposition are punished by not receiving food packs and cash transfers (see Eadie and Yacub, 2023). In our fieldwork, we heard people complaining about being ‘stricken off the list’ for cash grants for unknown reasons or for being deprioritised in reconstruction initiatives because they do not come from ‘most affected’ areas. Selective lists are corrosive to the community, creating resource envy for those not selected (Combinido and Ong, 2017).

People’s lived experiences of mistrusting lists extended to our deliberative forum. This was most evident when a community member confronted one of our team members about the participant selection process. The person approached our team member in the middle of the road, raised their voice, and accused us and the community leader of bias for only selecting 10 people to be part of the deliberative forum. They accused the community leader of manipulating the list of invited participants by ‘picking favourites’ and excluding people who had unpleasant encounters with the community leader. They criticised our team for the lack of transparency in participant selection. They suggested that next time, random selection should be conducted in public for everyone to see, like drawing names of participants from a raffle drum, to prove fairness in selection. Or, better yet, invite everyone to join the forum because selective participation is not only suspicious to the rest of the community but also offensive to those who do not get picked, given the history of politicised lists in aid distribution. Apart from emphasising the importance of transparency in random selection, this incident demonstrates that the credibility of sortition is hinged on people’s lived experiences of how population registers are used.

Unlike the deliberative forum in Marawi, the deliberative forum in Quezon City started with a profoundly problematic encounter with demographic information. We visited the headquarters of the local village council (barangay) to request a register of households that we can use as the basis of sortition. Such a list, we were told, does not exist. Instead, what the village official offered was a list of households marked as ‘clean’ from illegal drugs and households harbouring suspected drug dealers. At that time, policies related to the drug war were just being rolled out. The Department of Interior and Local Government launched a ‘community-based initiative’ that installed ‘drop boxes’ where community members placed anonymous tips of suspected drug dealers. This created a climate of distrust and fear within the community, such that the term list or listahan became synonymous with a ‘kill list’ or a register of people to be targeted in police operations or by vigilante death squads (see Rafael, 2021).

Recruiting participants in this context was a challenging task. We decided not to use a register of residents to select participants representative of the broader population. Instead, we decided to be purposive in our recruitment and put together a group of residents with different experiences of the drug war, regardless of their demographic profile. We shifted our priority from getting a democratic mirror to curating a safe space for meaningful conversation that can facilitate community healing (see Dryzek, 2005). We used sortition selectively. To invite people who had positive experiences with the drug war (e.g. those who think that their community felt safer), we went door to door to deliver invitations for households to nominate a person to join the forum. We ended up with roughly 20 responses. Of the 20, we randomly selected 7 respondents, who were then invited to join the forum. Meanwhile, we invited people who had a negative experience of the drug war by collaborating with a local organisation that provided psychological support to mothers and widows of men killed in the drug war. We have met many of these women in our field visits and have developed a relationship of trust over 3 years. The organisation picked six mothers and widows for our forum. We did not use sortition to select these participants. We recognised the psychologically charged meaning of being ‘selected from a list’. For survivors of the drug war, being selected from a list evokes terrifying memories of police officers identifying people from a drug watch list and inviting them ‘for a conversation’ where men would be made to swear to quit using illegal drugs. In some cases, these ‘conversations’ escalate into violence. Ultimately, our deliberative forum in Quezon City did not serve as a democratic mirror of the wider community, but it brought together a group of people with deep stories to share about the drug war.

Like the case of Marawi, the case of Quezon City unmasked our taken-for-granted assumptions about sources of data on which sortition can be based. In the context of state-sponsored violence, demographic data or ‘lists’ are tools for social control or, to use Michel Foucault’s (1978) term, an instrument of biopower that controls who lives and who dies. Critical demographers have advocated producing demographic data with and for marginalised communities serving the goals of emancipation instead of domination (Ortega, 2023), but until then, advocates of DMPs need to be cognisant of the politics of lists and the harms they cause that sortition-based selection could reinforce.

Negotiating the temporal demands of deliberation

Rigorous is how DMPs are often described. Compared to ‘faster’ forms of political participation like voting, signing a petition, or taking part in a town hall meeting, DMPs, the argument goes, give people the time to peruse expert evidence, scrutinise arguments, and reflect on a variety of views before collectively generating considered recommendations. Typically, DMPs last at least 4 days, so participants have sufficient time to learn and deliberate. Others, like the French Citizens’ Convention on Climate, took place over seven 3-day weekends (see Chrisafis, 2020).

As process designers, we made several assumptions to design a long form of citizen deliberation. We assumed that people have control over their time. We assumed that they have temporal flexibility such that they take leave from work and arrange caring duties or that they have predictable schedules so they can make plans to join the deliberations. We also assumed that participants had physical mobility so they could travel to the venue of the forum. We assumed that the cost of participation could be reduced by financially compensating participants, providing transportation, or even offering childcare support. These assumptions did not hold.

Our team originally planned for a 2-day deliberative forum for both Marawi and Quezon City. While 2 days may not be the ‘gold standard’ for DMPs, this could afford us time to execute a reasonably paced process plan. But even 2 days were too demanding for potential participants. Many hesitated to commit more than a day. In Quezon City, several participants who agreed to participate for 2 days started dropping out closer to the forum.

The contingencies of everyday life shaped our participants’ refusal to participate in a civic exercise that demanded much of their time and attention. Their concern, we surmised, was less about the high cost of participation. If this were the case, then participants would have been incentivised to join, given the generous compensation (higher than minimum wage) we offered for their participation, as well as the organised transportation to and from the venue of the forum. In Quezon City, we met a mother and grandmother who insisted on bringing their children and grandchildren to the deliberative forum, even though they could, in theory, ask their neighbours or relatives to take over childcare duties or to request our team to commission daycare providers. They had a clear preference to spend time with the children and not, for a second, keep the children out of sight. Both women just lost their husband and son to violent police operations. They considered it their personal responsibility to keep these men’s children out of danger. A day of deliberation was manageable – a day away that could be stimulating change for children – but an extended period away from home, spent in a conference room, would have disrupted the rhythm of everyday life that could be psychologically burdensome. Meanwhile, for other participants, 2 days of deliberation meant foregoing opportunities to work. Some participants worked in the ‘hustle economy’ where work rights such as taking paid or unpaid leave did not exist. The cost of participation in a deliberative forum was not the loss of income. The cost of participation was their reputation. In the hustle economy – whether it involved reselling scrap metal, unloading cement from trucks, or sewing garments – a good worker is a worker who shows up when summoned, even at the last minute. Declining work, even for a good reason, blemishes their record. They are seen as undependable or choosy, undeserving of being invited for future jobs.

These examples, among others, make it clear that when it comes to long-form deliberation, there is a ‘non-substitutability of time and money’ (Elliot, 2023: 66). We, as process designers, may appeal to people’s sense of civic duty, but in that process, we are laying claim to time that people wish to allocate to other meaningful life projects. For some critics of DMPs, this is the fundamental flaw of this democratic innovation – that a demanding form of political participation infringes people’s moral right to control their time (Elliot, 2023; also see Talisse, 2019). To put this argument more forcefully, the question is not whether long-form deliberation is possible. The question is whether it is even normatively desirable from the perspective of temporal justice.

Aside from temporal concerns, long-form deliberation also makes assumptions about the physical space in which the forum occurs. It is often taken for granted that a deliberative forum happens in a context of safety – a place where people can speak their minds and, on a more fundamental level, a place where their physical security is not at risk. We decided to host both deliberative forums on university campuses. In both contexts, universities are considered physically and intellectually safe spaces. We booked comfortable but modest rooms not to intimidate participants who have not set foot in a university before. In Marawi, we made a courtesy call to military officers to inform them of our activity. At that time, Marawi was still under Martial Law, so we closely monitored security reports about potential terrorist threats. In Quezon City, we informed village leaders that we were organising a non-partisan forum at a university close by. At that time, the Philippines was turning in an illiberal direction, with activist and opposition groups increasingly facing pressure from the government. We emphasised the non-partisan nature of our forum so as not to cause harm to our participants. We were conscious to end the forum before sunset. In both contexts, travelling home in the dark could be dangerous. We booked a minibus that shuttled participants to and from their communities to reduce the risk of participants travelling alone. While it was fruitful for participants to be taken out of their everyday contexts to reflect and speak about their experiences, replicating conditions that ensure physical safety is an increasingly challenging, if not irresponsible, enterprise as deliberations take more days. Designing a hybrid long-form deliberation was not an option at that point either. Marawi was a ‘digital desert’, while participants in Quezon City already spoke about the drug war in hushed tones in their homes.

In the end, we settled for a day-long deliberation. Both forums followed a three-part format of (1) norm-setting, where participants themselves established the rules of discussion; (2) problem identification, where participants co-constructed a definition of their community’s concerns; and (3) charting plans of action, where participants deliberated on the barriers and openings for collective problem-solving. The remit of deliberation in both forums was relatively open ended compared to DMPs designed to inform policymaking. In Marawi, we asked participants what people-led recovery meant to them and how the recovery process can be improved. The term ‘people-led recovery’ was often used by government and aid agencies as well as civil society groups, and so our forum aimed to ground its meaning on the everyday life of displaced communities. In Quezon City, we asked participants to describe the security issues they faced in their community and how they can be addressed. We refrained from using the terms ‘drug war’ or ‘police brutality’ so as not to intimidate participants. We used a combination of plenary sessions and facilitated breakout groups. We did not invite experts to provide evidence and instead prioritised participants’ own expertise based on their lived experience. Evidently, the quality of discussions was not as epistemically rigorous as in British Columbia, where Assembly Members deliberated on the technicalities of electoral laws, or in Ireland, where the theological and moral claims about abortion were examined. Our deliberative forums were humbler, but they were nevertheless as meaningful. They created spaces for everyday people to express their grievances and transform community relationships.

In Marawi, it became clear that participants found the imposition of Martial Law legitimate if this meant swifter government action and security from terrorist threats. But even under military rule, internally displaced communities demanded robust mechanisms for public consultation and accountability, especially on the issue of reconstruction. Their main concern was to connect their grievances to authorities.

In Quezon City, our deliberative forum served as a space for community healing. The forum started tense, with supporters of the drug war accusing mothers and wives of suspected drug dealers of being irresponsible because they did nothing to stop their sons and husbands from selling drugs to young people in the village. However, as participants focused on how security can be improved in their community, they developed empathy for each other. They realised they were facing the same problem as parents of children at risk of being recruited into the drug trade. The forum ended in tears. The mothers and widows of men killed in the drug war apologised to their neighbours. They admitted that their sons’ and husbands’ involvement in the drug trade put their neighbours’ children at risk, and for that, they are sorry. Meanwhile, supporters of the drug war assured the mothers and widows that they could still be proud of their sons and husbands, for they did everything they could to provide for their families. This deliberative forum may not have served the role of epistemic filter that generated robust recommendations that can inform policy, but it demonstrated the power of deliberation in empowering people to be the authors of their own community’s narratives and build horizontal relationships, instead of being defined by the divisive and dehumanising rhetoric of the populist strongman in power. This is what a day-long, short-form deliberation can offer.

Conclusion

In this article, we put forward our reflections on designing citizen deliberation in fragile contexts. We made a case for unmasking our taken-for-granted assumptions about the core design features of DMPs and, in so doing, recognise the potential harms of sortition and long-form deliberation when we apply them in unsuitable contexts. We conclude our article by offering two lessons we can learn from our empirical work.

First, the theory and practice of democratic deliberation need to problematise their assumptions about sortition systematically. In our fieldwork, we characterised the problematic sources of demographic information on which sortition could be based, as well as the ways in which population registers are weaponised to intimidate, control, and exclude populations. Population registers and demographic information are political constructs. We argued that the credibility of sortition is only as good as people’s lived experiences of how these registers are used. This insight is not unique to the Philippines. Scholars of indigenous demography have long interrogated the statistical construction of indigenous populations and how these statistical constructions have historically been used as instruments of exclusion and colonial domination (see Kukutai and Taylor, 2016), while scholars of critical demography have criticised the ways in which ‘White logic’ and ‘White methods’ create the illusion of ‘objective’ demographic data from which we base the design of democratic innovations (see Zuberi and Bonilla-Silva, 2008). DMPs cannot be an emancipatory project if its design is built on exclusionary and colonial foundations.

Second, future research is needed to uncover the implications of long-form deliberation in democratic exclusion. In this article, we interrogated the kind of political agent we assume to be willing and able to commit to long-form deliberation and the dangers of these assumptions in practice. While it is heart-warming to hear stories of bus drivers, farmers, nurses, refugees, and other everyday people taking part in citizens’ assemblies, we need to listen out to stories and justifications of everyday people who refuse to take part in long-form deliberation. Advocates of DMPs argue that sortition encourages participation of people beyond ‘the usual suspects’. It is time to understand why people who do not participate are also ‘the usual suspects’. Indeed, there is no winning formula for citizen deliberation, not even the ‘flavour of the month’.